Re: [PATCH] Add ALWAYS DEFERRED option for constraints

On Fri, Nov 03, 2017 at 01:41:45PM -0400, Peter Eisentraut wrote:
> On 11/2/17 16:54, Nico Williams wrote:
> > Replacing condeferred and condeferrable with a char columns also
> > occurred to me, and though I assume backwards-incompatible changes to
> > pg_catalog tables are fair game, I assumed everyone would prefer
> > avoiding such changes where possible.
>
> I don't think there is an overriding mandate to avoid such catalog
> changes. Consider old clients that don't know about your new column.
> They might look at the catalog entries and derive information about a
> constraint, not being aware that there is additional information in
> another column that overrides that. So in such cases it's arguably
> better to make a break.

Makes sense.

> (In any case, it might be worth waiting for a review of the rest of the
> patch before taking on a significant rewrite of the catalog structures.)

I'll wait then :)

When you're done with that I'll make this change (replacing those three
bool columns with a single char column).

> > Hmmm, must I do anything special about _downgrades_? Does PostgreSQL
> > support downgrades?
>
> no

Oh good. Thanks for clarifying that.

Nico
--